A Novel Non-invasive Model Based on GPR for the Prediction of Liver Fibrosis in Patients With Chronic Hepatitis B

Background: Some controversy remains regarding conventional serum indices for the evaluation of liver fibrosis. Therefore, we aimed to combine the existing index with other serum parameters to discriminate liver fibrosis stages in patients with chronic hepatitis B (CHB). Methods: A total of 1,622 treatment-naïve CHB patients were divided into training (n = 1,211) and validation (n = 451) cohorts. Liver histology was assessed according to the Scheuer scoring scheme. All common demographic and clinical parameters were analyzed. Results: By utilizing the results of the logistic regression analysis, we developed a novel index, the product of GPR, international normalized ratio (INR), and type IV collagen (GIVPR), to discriminate liver fibrosis. In the training group, the areas under the ROCs (AUROCs) of GIVPR, APRI, FIB-4, and GPR for significant fibrosis were 0.81, 0.75, 0.72, and 0.77, respectively; the AUROCs of GIVPR, APRI, FIB-4, and GPR for advanced fibrosis were 0.82, 0.74, 0.74, and 0.78, respectively; and the AUROCs of GIVPR, APRI, FIB-4, and GPR for cirrhosis were 0.87, 0.78, 0.78, and 0.83, respectively. Similar results were also obtained in the validation group. Furthermore, the decision curve analysis suggested that GIVPR represented superior clinical benefits in both independent cohorts. Conclusion: The GIVPR constructed on GPR represents a superior predictive model for discriminating liver fibrosis in CHB patients.


BACKGROUND
Hepatitis B virus (HBV) infection is a serious public health problem. It is estimated that more than 350 million people are chronically infected worldwide (1). From 1990 to 2013, the mortality rate of liver cirrhosis and hepatocellular carcinoma caused by HBV infection increased by 33% worldwide (2). Based on the outcomes of patients who receive early diagnosis and effective antiviral therapy, the prognosis of CHB can be significantly improved even if the case is histologically advanced fibrosis or cirrhosis (3). Therefore, it is of great importance to assess the risk of early liver fibrosis in CHB patients.
Currently, the gold standard for the assessment of liver fibrosis is still liver biopsy. However, its limitations, such as its invasiveness, sampling errors, cost, intra-and interobserver discrepancies, and the risk of potentially life-threatening complications, restrict its clinical application (4). Clinical practice requires simple operations or non-invasive and easy methods to diagnose liver inflammation, injury or fibrosis (5). The World Health Organization (WHO) guidelines recommend serologic biomarkers and FibroScan as useful non-invasive methods for evaluating CHB patients (6). However, several factors, including necroinflammatory activity, ascites, cost, and lack of skilled operators, may diminish the clinical use of FibroScan (6,7). Serum biomarkers are particularly important in these methods because they do not require qualified staff and expensive equipment for evaluation (8). The WHO has recommended the aspartate aminotransferase (AST)-platelet ratio index (APRI) and fibrosis-4 (FIB-4) as non-invasive indices for CHB patients (6). The diagnostic value of these two indices in liver fibrosis has been widely studied, but their sensitivity and specificity are still controversial (9). Recently, a study by Lemonie et al. (10) suggested that the γ-glutamyl transpeptidase to platelet ratio (GPR) was more accurate than APRI or FIB-4, and this study was supported by several studies on Chinese subjects (11,12). However, there were still a few inconsistent conclusions (13). Therefore, novel non-invasive serum calculations are still needed because the current biochemical markers do not have enough diagnostic accuracy to replace liver biopsy.
Serum collagen, especially type IV collagen, has been confirmed to be a useful, non-invasive marker for measuring the activity of this pathway at a single time point and has been shown to reflect prognosis and responses to a variety of chronic liver diseases (14). INR is a routine serological marker associated with liver function and essentially reflects the progression of liver diseases. Wu et al. reported that the INR was an independent factor for the prediction of significant fibrosis in patients with CHB (6,15).
More efforts should be dedicated to pursuing simple, safe and reliable non-invasive diagnostic measures to stage liver fibrosis. In this study, we aimed to construct and validate a predictive index consisting of GPR, INR, and type IV collagen to reflect liver fibrosis simply and effectively in CHB patients.

Patients
Overall, between January 2014 and January 2021, we retrospectively screened 2,193 consecutive Chinese individuals with chronic hepatitis B who underwent liver biopsy and clinical examination at Shanghai Public Health Clinical Center, Fudan University. CHB was diagnosed when serum hepatitis B surface antigen (HBsAg) was persistently positive for more than 6 months (16). All the patients were >18 years old. Non-alcoholic fatty liver disease (NAFLD) was diagnosed as at least 5% biopsy-proven hepatic steatosis without significant alcohol consumption (17). The exclusion criteria were as follows: antiviral treatment history, coinfection with hepatitis C virus (HCV), hepatitis D virus (HDV), hepatitis E virus (HEV), or human immunodeficiency virus (HIV), significant alcohol consumption (>20 g/d), autoimmune hepatitis, hepatocellular carcinoma, decompensated cirrhosis, inadequate liver biopsy samples (<1.5 cm), and the use of warfarin.
We summarized the flow diagram of the study population in Figure 1. After excluding patients with coinfection with HCV, HDV, HEV, or HIV (n = 113), alcohol consumption (>20 g/d) (n = 104), autoimmune hepatitis (n = 51), history of antiviral treatment (n = 128), and incomplete clinical data (n = 79), 1,662 treatment-naïve patients with CHB were included. The population was randomly divided into a training set (n = 1,211) and a validation set (n = 451) for model development and validation using SPSS software.

Liver Biopsy
Percutaneous liver biopsy was performed using a 16 G needle under ultrasound guidance. Liver samples with a minimum length of 1.5 cm and at least 7 complete portal tracts were fixed in 10% formalin, embedded in paraffin, and stained with HE Masson's trichrome and reticulin for histological analysis. Liver histology was analyzed by two experienced pathologists who were blinded to other clinical and laboratory data and classified according to the Scheuer scoring system (18) as follows: S0 (no fibrosis), S1 (mild fibrosis without septa), S2 (moderate fibrosis with few septa), S3 (severe fibrosis with numerous septa without cirrhosis), and S4 (cirrhosis). In this study, liver fibrosis stage ≥S2 was defined as significant fibrosis, ≥S3 was defined as advanced fibrosis, and S4 was defined as cirrhosis. These definitions represent at minimum significant fibrosis and affect the management of patients in terms of treatment indications (16,19).

Laboratory Data
Fasting blood samples were obtained within a week of liver biopsy. Platelets and other blood cells were counted using a Sysmex-XT 4000i automated hematology analyzer. The international normalized ratio (INR) and other coagulation indices were measured using a STAR Max automatic coagulation analyzer. Alanine transaminase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), γ-glutamyl transferase (GGT), hyaluronic acid, laminin, N-terminal propeptide of type III procollagen (PIIINP), type IV collagen, and other serum biochemical parameters were measured using an Architect C16000 automatic biochemical analysis system.

Statistical Analysis
Statistical analysis was performed using IBM SPSS Statistics version 26.0 (SPSS Inc., Chicago, USA) and R 4.0.3 (http://www. R-project.org). Continuous variables are expressed as the mean ± standard deviation or median (interquartile range, IQR) and were compared using Student's t-test (for normally distributed continuous variables) or the independent Mann-Whitney U-test (for non-normally distributed continuous variables). Categorical variables are expressed as proportions and were compared by the chi-square test. Logistic regression models were used to assess the correlations between variables and liver fibrosis. The performances of the non-invasive markers for predicting liver fibrosis were assessed by receiver operating characteristic (ROC) curve analyses. The Delong Z-test was used to compare the AUROCs of the serum models. Decision curve analysis (DCA) was used to further evaluate the predictive performances. A two-sided P < 0.05 was considered statistically significant.
Similarly, in the validation set, compared to the other four serum indices, GIVPR had the highest AUCs of 0.82 (sensitivity 73.82% and specificity 75.23%) for predicting significant fibrosis, 0.85 (sensitivity 81.67% and specificity 70.09%) for predicting advanced fibrosis, and 0.80 (sensitivity 84.42% and specificity 78.88%) for predicting cirrhosis ( Table 4). These results suggest that GIVPR is an excellent predictor of liver fibrosis in CHB patients.

DCA for the Clinical Utility of GIVPR
Moreover, we conducted DCA to further investigate the clinical application value of GIVPR, GIVPTAR, APRI, FIB-4, and GPR for predicting liver fibrosis. In the training group, DCAs revealed that from a threshold probability of 20-80%, the application of GIVPR to predict liver fibrosis risk increased the benefit considerably more than the other four scores (Figure 5). Regarding the validation group, the DCAs of GIVPR also showed a better net benefit with a wide range of threshold probabilities and better performances for predicting liver fibrosis than GIVPTAR, APRI, FIB-4, and GPR (Figure 6).

DISCUSSION
Early diagnosis and accuracy in evaluating liver fibrosis or cirrhosis may play important roles not only in controlling disease progression but also in the treatment of chronic HBV infection (22). Liver biopsy is the gold standard for evaluating liver fibrosis in chronic liver disease. However, although liver biopsy is usually a safe procedure, it has some technical limitations and risks (23). Thus, there is an increasing need for simple and reliable non-invasive predictors for liver fibrosis, some  of which have been evaluated in multiple studies. However, how their sensitivity and accuracy are affected by various factors is still a matter of debate (24). By combining noninvasive indicators, the overall diagnostic coincidence rate can be improved.
In the present study, we assessed the relationships between serum parameters and non-invasive indices and liver fibrosis in CHB patients. GIVPR and GIVPTAR based on GPR all exhibited excellent capacities to predict the progression of liver fibrosis. However, GIVPTAR, which required more variables, did not obtain higher AUCs than GIVPR and did not improve the predictive performance for liver fibrosis. We also compared the predictive accuracy of GIVPR with APRI, FIB-4, and GPR. Our results showed that in both the training and validation cohorts, FIGURE 5 | Liver fibrosis decision curve analysis in training set. Decision curve analysis depict the clinical net benefit. GIVPR is compared with GIVPTAR, APRI, FIB-4, and GPR for predicting significant fibrosis (A); GIVPR is compared with GIVPTAR, APRI, FIB-4, and GPR for predicting advanced fibrosis (B); GIVPR is compared with GIVPTAR, APRI, FIB-4, and GPR for predicting cirrhosis (C). Black line, net benefit when no patient will experience the event; gray line, net benefit when all patients will experience the event. The preferred markers is the marker with the highest net benefit at any given threshold. FIGURE 6 | Liver fibrosis decision curve analysis in validation set. Decision curve analysis depict the clinical net benefit. GIVPR is compared with GIVPTAR, APRI, FIB-4, and GPR for predicting significant fibrosis (A); GIVPR is compared with GIVPTAR, APRI, FIB-4, and GPR for predicting advanced fibrosis (B); GIVPR is compared with GIVPTAR, APRI, FIB-4, and GPR for predicting cirrhosis (C). Black line, net benefit when no patient will experience the event; gray line, net benefit when all patients will experience the event. The preferred markers is the marker with the highest net benefit at any given threshold.
GIVPR had the best AUC value for staging significant fibrosis, advanced fibrosis, and cirrhosis. Thus, GIVPR, which requires only GPR, INR, and type IV collagen and is simple to calculate, has a more powerful predictive performance for liver fibrosis in CHB patients.
There were two kinds of serum biomarkers for liver fibrosis progression, indirect serum markers and direct serum markers (25). Indirect serum markers had no direct correlation with liver fibrosis but reflected liver dysfunction or other fibrosis-related symptoms. They are often calculated into mathematical formulas or may be used individually (26). APRI and FIB-4 are the two non-invasive procedures for evaluating liver fibrosis that receive the most attention. They were reported to have a high AUROC to detect significant fibrosis and cirrhosis in CHB patients in East Africa and Asia (27,28). The WHO CHB guidelines also recommend APRI and FIB-4 for application in resource-limited health care regions (29). However, a meta-analysis suggested that their diagnostic performance was not good enough to discriminate liver fibrosis in CHB patients and could not be used as an ideal replacement for liver biopsy (30). GPR is a novel index to assess liver fibrosis in patients with CHB in West African cohorts. It was shown to be better than the classical models APRI and FIB-4 (10). Additionally, GPR was reported to diagnose significant liver fibrosis and cirrhosis well in a large cohort of HBV monoinfected Gambian patients using FibroScan measures as a reference (31). However, GPR showed a less clear advantage in a Brazilian cohort and other Chinese cohorts (13,32). In this study, our GIVPR model showed acceptable distinguishing power for the prediction of significant live fibrosis, advanced liver fibrosis, and cirrhosis in the training set, with AUCs of 0.797, 0.815, and 0.844, respectively; similar results were obtained in the validation set. Furthermore, we confirmed significantly better performance for the assessment of liver histological scores compared to the biochemical marker panels APRI, FIB-4, and GPR. Due to the different inflammatory and clinical conditions of patients with chronic hepatitis B and chronic hepatitis C, the effect of etiology on fibrosis progression and clinical biomarkers can explain this result (33,34).
Moreover, the indirect serum markers evaluated in this study included the measurement of coagulation parameters, which were found to increase with the progression of liver fibrosis. Among these routine markers, INR was identified as an independent factor for the prediction of significant fibrosis and cirrhosis in CHB patients. Sterling et al. (21) reported that the INR was an independent predictor of liver fibrosis, and its concentration was directly related to liver function. Another study demonstrated that the INR level was associated with liver fibrosis and used INR as a parameter in their King's score, which was closely related to the progression of liver fibrosis (35,36).
Direct biomarkers of liver fibrosis are fragments of liver matrix components produced in the process of fibrosis. These markers represent the intensity of fibrogenesis or fibrinolysis, such as type IV collagen, laminin, hyaluronic acid and metalloproteinases (37). Serum collagen levels, especially type IV, have been shown to be a useful, non-invasive measure of the activity of this pathway at a single time point and have been shown to reflect prognosis and responses to a variety of chronic liver diseases (14). Type IV collagen is an important component of the normal extracellular matrix. Compared with type I and type III collagen, which are partially hydrolyzed, type IV collagen remains intact in the matrix; therefore, the serum composition of type IV collagen is considered to mainly reflect the degradation of the matrix (38). Serum type IV collagen has been confirmed to be associated with both the progression of liver inflammation and fibrosis, which is in line with our data (26,39).
This study has several limitations worth considering. First, this was a retrospective study in a single center and should be further confirmed in more patients from other centers. Second, GIVPR was not dynamically observed. We recommend further investigation into the efficacy of GIVPR compared to other non-invasive indices in evaluating fibrosis progression and in predicting liver-related end-stage disease after long-term antiviral inhibition of HBV.

CONCLUSION
In summary, a novel non-invasive calculation, GIVPR, was established from GPR, INR, and type IV collagen. GIVPR demonstrates superior diagnostic accuracy and clinical usefulness compared to conventional serum indices. Although the clinical usefulness of GIVPR warrants future investigation, our findings showing that GIVPR is non-invasive and easily administered indicate that it could be a promising tool for the discrimination of liver fibrosis, especially in resource-limited regions.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.